The field of natural language processing is witnessing a significant shift towards more efficient and flexible tokenization methods and vocabulary management strategies. Researchers are exploring new approaches to address the limitations of traditional tokenization methods, such as Byte Pair Encoding (BPE), and to improve the representational power of language models. One notable direction is the development of dynamic and hierarchical tokenization methods that can capture rare words and out-of-vocabulary tokens more effectively. Another area of focus is the redesign of vocabulary structures, with a move towards more compositional and compact representations that can leverage the underlying structure of language. These advancements have the potential to improve the performance and generalizability of language models, particularly in multilingual and low-resource settings. Noteworthy papers in this area include: Vocab Diet, which proposes a compact reshaping of the vocabulary using vector arithmetic, and DVAGen, which introduces a unified framework for dynamic vocabulary-augmented language models. These papers demonstrate innovative solutions to long-standing challenges in language model tokenization and vocabulary management, and are likely to have a significant impact on the field.